
% Table generated with cell2latex.m  on  13-Jan-2022 06:48:28

\begin{table}[H]
\captionsetup{justification=centering}
\centering
\caption{WhatsApp RD and RCT Results -- 2021 \label{tab:RCT2021}}
\begin{adjustbox}{max width=\textwidth,max totalheight=\textheight-14\baselineskip}
%\footnotesize
\begin{tabular}{lrrrrrrrrr}
\addlinespace
\toprule
  &\multicolumn{1}{c}{(1)} &\multicolumn{1}{c}{(2)} &\multicolumn{1}{c}{(3)} &\multicolumn{1}{c}{(4)} &\multicolumn{1}{c}{ } &\multicolumn{1}{c}{(5)} &\multicolumn{1}{c}{(6)} &\multicolumn{1}{c}{(7)} &\multicolumn{1}{c}{(8)}\\
 &\multicolumn{4}{c}{General risk information} &\multicolumn{1}{c}{ }&\multicolumn{4}{c}{Personalized risk information}\\\cline{2-5}\cline{7-10}
 &\multicolumn{2}{c}{RCT}&\multicolumn{2}{c}{RD} &\multicolumn{1}{c}{ }&\multicolumn{2}{c}{RCT}&\multicolumn{2}{c}{RD relative to}\\\cline{2-5}\cline{7-10}
  &\multicolumn{1}{c}{} &\multicolumn{1}{c}{} &\multicolumn{1}{c}{} &\multicolumn{1}{c}{} &\multicolumn{1}{c}{ } &\multicolumn{1}{c}{} &\multicolumn{1}{c}{}&\multicolumn{2}{c}{general risk treatm.}\\\cline{9-10}
  &\multicolumn{1}{c}{ITT} &\multicolumn{1}{c}{IV} &\multicolumn{1}{c}{ITT} &\multicolumn{1}{c}{IV} &\multicolumn{1}{c}{ } &\multicolumn{1}{c}{ITT} &\multicolumn{1}{c}{IV} &\multicolumn{1}{c}{ITT} &\multicolumn{1}{c}{IV}\\\midrule

\\
\multicolumn{10}{l}{\textit{A. Balance}} \\
Economically Vulnerable  & 0.001 &  & -0.013 &  &   & 0.003 &  & -0.008 &  \\ 
  & (0.015) &  & (0.021) &  &   & (0.009) &  & (0.016) &  \\ 
Rural  & -0.007 &  & 0.007 &  &   & 0.001 &  & 0.006 &  \\ 
  & (0.004) &  & (0.005) &  &   & (0.002) &  & (0.004) &  \\ 
\\
\multicolumn{10}{l}{\textit{B. Message receipt}} \\
WhatsApp read  & 0.793 &  & 0.798 &  &   & 0.792 &  & 0.005 &  \\ 
  & (0.005) &  & (0.012) &  &   & (0.003) &  & (0.014) &  \\ 
Report card intention to treat  & 0.004 &  & 0.013 &  &   & 0.008 &  & 0.026 &  \\ 
  & (0.014) &  & (0.019) &  &   & (0.008) &  & (0.015) &  \\ 
\\
\multicolumn{10}{l}{\textit{C. Outcomes in clean 5 days before report card}} \\
Any modification  & 0.041 &  & 0.043 &  &   & 0.081 &  & 0.037 &  \\ 
  & (0.003) &  & (0.007) &  &   & (0.003) &  & (0.008) &  \\ 
Add any  & 0.039 &  & 0.038 &  &   & 0.080 &  & 0.038 &  \\ 
  & (0.003) &  & (0.006) &  &   & (0.003) &  & (0.008) &  \\ 
Schools Added  & 0.070 & 1.807 & 0.067 & 1.766 &   & 0.156 & 1.957 & 0.086 & 2.013 \\ 
  & (0.007) & (0.114) & (0.015) & (0.257) &   & (0.006) & (0.052) & (0.020) & (0.441) \\ 
$\Delta$ Risk  & -0.002 & -0.050 & -0.001 & -0.024 &   & -0.013 & -0.157 & -0.000 & -0.057 \\ 
  & (0.001) & (0.012) & (0.001) & (0.035) &   & (0.001) & (0.008) & (0.002) & (0.058) \\ 
\\
\multicolumn{10}{l}{\textit{D. Endline outcomes}} \\
Any modification  & 0.054 &  & 0.053 &  &   & 0.099 &  & 0.055 &  \\ 
  & (0.005) &  & (0.009) &  &   & (0.004) &  & (0.010) &  \\ 
Add any  & 0.051 &  & 0.046 &  &   & 0.096 &  & 0.055 &  \\ 
  & (0.005) &  & (0.009) &  &   & (0.004) &  & (0.009) &  \\ 
Schools Added  & 0.098 & 1.921 & 0.085 & 1.856 &   & 0.203 & 2.108 & 0.148 & 2.978 \\ 
  & (0.015) & (0.236) & (0.022) & (0.313) &   & (0.010) & (0.067) & (0.025) & (0.611) \\ 
$\Delta$ Risk  & -0.001 & -0.029 & -0.002 & -0.034 &   & -0.016 & -0.163 & -0.004 & -0.149 \\ 
  & (0.001) & (0.017) & (0.002) & (0.049) &   & (0.001) & (0.009) & (0.003) & (0.061) \\ 
\bottomrule
%\addlinespace
\end{tabular}
\captionsetup{justification=justified}
\floatfoot{\scriptsize
\textit{Notes. }ITT and IV effects of 2021 WhatsApp warnings intervention. RCT columns (1) and (2): effects of random assignment to general risk information message vs. control group for students with predicted risk $\in (.2,.3]$. RCT columns (5) and (6): effects of random assignment to personalized risk information message vs. control group for students with predicted risk $>.3$. Robust SEs in parentheses. General information intervention N=6,819.  Personalized information intervention N=18,763. RD columns (3) and (4): regression discontinuity evaluation of general risk information message vs no treatment around 0.20 cutoff. RD columns (7) and (8): regression discontinuity evaluation of personalized risk information message vs general risk information message around 0.30 cutoff. RD specifications computed using local linear fit with a bandwidth of 0.1. Standard errors are heteroskedasticity-robust nearest neighbor variance estimator with minimum of 3 neighbors, as in \citet{calonico2014robust}. ITT column shows effects of group assignment. IV columns show the instrumental variable specification, where the endogenous regressor is the add any school indicator, instrumented with group assignment for the RCT, and with a dummy of crossing the risky threshold for the RD. Panel A: balance tests on predetermined characteristics. Panel B: message receipt. ``WhatsApp read'' is an indicator equal to one if applicant views the WhatsApp treatment message. ``Report card intention to treat'' is indicator for receiving a report card 5 days later. Report card included additional information on nonplacement probability and school options. 28\% of the RCT sample were assigned to received the report card. Panel C: outcomes within 5 days window between WhatsApp intervention and report card was sent.  Panel D: endline choice behavior. See the description of the 2021 intervention in section \ref{app:int} for details.  
}
\end{adjustbox}
\end{table}
